I am trying to produce mutltiple tables using the flextable and kable packages. When I want to output some table iteratively, I found knit_print
of flextable is not working in loop. Below is a minimal example:
---
output: word_document
---
```{r}
library(flextable)
library(knitr)
```
```{r}
data(cars)
speed<-unique(cars$speed)
for (v in 1:length(speed)) {
carspd<-cars[which(cars$speed==speed[v]),]
tb<-regulartable(carspd)
knit_print(tb)
}
knit_print(tb)
```
Just the last knit_print
can print the result to the word_document with the .Rmd
file.
Now I find the difference of them in .md
which is output by the pandoc process file with ultraedit,
the right table:
```{=openxml}
<w:tbl xmlns:w=".......
the wrong table:
鈥媊``{=openxml}
<w:tbl xmlns:w="
In hexadecimal there are extra content:"E2 80 8B", someone call they Zero-Width Space? But I am not figure out how to avoid it.
This is a limitation of one of the functions that flextable
uses in its knit_print
method: knitr::asis_output()
can't be used in a loop. A couple of solutions are described here: https://github.com/yihui/knitr/issues/1137.
This one works for me in HTML output, just collecting all the output into one big vector and printing it at the end. I have modified the code to also insert some text before the table and a figure after it. I don't use Word, but I'd assume it will also work there:
---
output: html_document
---
```{r}
library(flextable)
library(knitr)
```
```{r}
data(cars)
speed <- unique(cars$speed)
results <- character()
for (v in 1:length(speed)) {
carspd <- cars[which(cars$speed == speed[v]),]
tb <- regulartable(carspd)
# Generate a figure in a temporary file
filename <- tempfile(fileext = ".png")
png(filename)
hist(carspd$dist)
dev.off()
# Put everything into the results vector
results <- c(results, "\n\nThis is the table for v =", v,
knit_print(tb),
knitr:::wrap(include_graphics(filename)))
}
asis_output(results)
knit_print(tb)
```
A couple of notes:
\n\n
in the text seems to be necessary to insert a line break.knitr:::wrap()
function is an undocumented internal function in knitr
, so there might be limitations that I don't know about, and this might fail in some future version of knitr
.